I want to preface this post by saying that I am extremely frustrated right now, but I'm trying to make this a solution based post, and stop this from happening to myself or others on this site. Sorry if it comes off a bit rant-ish though.

This problem has been discussed on Puzzling Meta, in this 2017 post as well as an earlier post by me in which I made some suggestions and is related to the meta questions What should I do if I've made a mistake in my question? and Downvoting answers made before question edits/clarifications. However, the problem has not been resolved, and I have a new set of suggestions, making this question not a duplicate.

The problem:

Yesterday was the second time that I have posted a correct answer followed by the questioner changing the question in a way that makes my answer invalid. It is extremely frustrating to work for hours on a solution for it to be invalidated, downvoted, not accepted, and commented on for not meeting the rules, because the questioner changed the rules.

I have only been on Puzzling Stack Exchange for a little over a year, and I post answers to maybe about 1 out of every 200 puzzles I see, which is not a lot. Yet this problem has happened to me twice already, and as seen by the various meta posts on the topic, happens to others as well, which worries me about how prevalent the issue is on this site.

My experiences:

The first occurrence was on the question One truthteller/liar and one random person. The question was to come up with a strategy to distinguish Alice (either a truthteller or liar) from Bob (a random). When I answered, the question explicitly stated that randoms randomly decide to lie or tell the truth on each question. Not long after, the question had been edited to say that a random randomly decides say yes or no on each question. When I objected I received this comment:

@ankit, my original intent was to say that the random says yes/no at random. I did not realise that there might be a difference between the random, randomly saying yes/no and the random, randomly saying the truth or lying. While I am still not 100% convinced if it makes any difference, I apologise for any confusion. Request you to not only leave your answer to the original question as it is but also to post a new answer to the question in it's correct format. – Hemant Agarwal Aug 28 '20 at 17:41

The second occurrence was yesterday on Secretly press different buttons. This question was about 2 button pressers, among a group of 12, trying to press different buttons, without being identified by 3 of the 12, who can win by identifying one button pressers. The puzzle stated that all communications must be done public, and that the button pressers had their devices with buttons from the start. My answer was to use DHP cryptography (a method where everything, including the creation of the public key, can be done while maintaining public communication that everyone can hear/see, while only the intended recipients are able to understand the message). I proved that my solution was optimal, therefore making it the correct answer. I got the following message before the question was changed, and my answer was invalidated:

Hah, you managed to use not one but two loopholes that were totally against the spirit of the problem: cryptography and knowledge of the buttons themselves. I adjusted the puzzle to close these. (I'm not the downvoter, btw.) – Joseph Sible-Reinstate Monica Aug 28 '21 at 21:53

Why this is the problem:

The first post had a comment that said that my answer went against "my [the questioner's] original intent". The second said my answer was "against the spirit of the puzzle."

But what is "my original intent?" What is "the spirit of the puzzle?" No one knows. It is not written down. With all due respect, both comments simply mean "I messed up and didn't intend to allow that." When solving puzzles, you are supposed to make a solution that works by the written rules of the puzzle. There should be no requirement to meet additional unwritten rules that are stuck in someone's head and inaccessible to the public.

To quote @Rubio from a previous meta post

"It is manifestly unfair to leave answers made prior to invalidating edits subject to downvotes by people unfamiliar with the history of the question, if the answer was a fitting answer to the question as it stood at the time the answer was made. For that reason, we strongly discourage such edits."

Previously proposed solutions:

  1. Last time I meta posted on this topic, I suggested requiring votes on edits after answers existed on the question. Honestly, my suggestions were extremely unrealistic. This time will be a bit more realistic-- I promise :)
  2. "I would suggest that if a question needs radical alteration, it should be posted as a separate question. If the first is so bad as to be unsalvageable, it should be closed/deleted by the OP. If the original question still works, it can be answered separately from the new version. – GentlePurpleRain Mod Jan 27 '17 at 15:57" (link to this comment). I think this is smart and in fact is what I requested in both my experience. However, I do not believe that simply suggesting it is enough, seeing as every occurrence documented in meta, it didn't work.
  3. "If a question is radically altered after posting, the fact should be explicitly acknowledged in the edited version of the question. - Gareth McCaughan Mod Jan 26 '17 at 17:28." (link to this quote). Sure but this is not enough. It still makes a right answer a wrong one, and quite often, people will skim through this and forget it by the time they are downvoting and commenting on pre-edit era questions.

Most of the previously proposed solutions I could find fall into these three categories.

My solution idea:

I suggest that an article be posted in Help Center where the rules are all posted titled "Questions with mistakes, loopholes or requiring clarifications" or something along those lines. This would set the rules regarding what to do in such situations, with the rules based on discussions following this post and the posts linked on the top on the same topic. Questions that get edited in a way that breaks rules in the article can then be flagged as being against Puzzling.SE Rules and closed. Once the edit is reverted to the original, the question can be reopened.

It honestly should be illegal by current rules via the Code of Conduct, as editing a problem like this is seriously discourteous, and pretty much the opposite of "respectful," "collaborative," and "as easy as possible for others to help you (answer the question)." But by explicitly making rules, it can be clearer what is expected and what is not allowed.

My specific beliefs on what rules should be:

  1. If the questioner makes any modifications for mistakes (excluding spelling/grammar/formatting), they must be explicitly written on the post, i.e.: "Edit 1: I accidentally wrote Bob instead of Alice in this sentence '...', Edit 2: I accidentally forgot to add this detail '...'," etc. In this way, the questioner is taking accountability for their own mistakes.
    • a. If the modification is made after an answer is posted, this should be indicated, ex: "Edit 3 (posted after @answerer 's answer) ___________"
  2. Any puzzle that is a valid puzzle (i.e. it has answers, it is possible, it makes sense) should be encouraged (but not yet enforced) not to be altered.
    • a. Should someone have an valid answer posted prior to the edit, and the edit invalidates the answer, the edit should not be allowed.
    • b. If someone posts an answer in a reasonable manner such that they were probably solving it before the edit and submitted the answer reasonably after the invalidating edit, the edit must be removed. Say I work 5 hours on a puzzle, and a few minutes before I post my solution, the answer is invalidated. That's not fair.
      • (i.) It is the responsibility of the answerer to tell the questioner that his/her answer is based on the previous version, why the edit invalidated it, and request that the question be reverted.
      • (ii) "Reasonable" as written in 2b is to be determined by the community (i.e. comments, chat discussions, meta posts, and close votes). If someone has posted 2 answers to other questions in between the edit and the answer to the edit question, it is clearly not reasonable. If the answer is posted a week after the edit, it is not reasonable. If it is within a few hours of the edit, it is almost definitely reasonable.
  3. Valid questions with mistakes should be encouraged to be re-asked (not deleting the original with mistakes) without the mistakes. These will not be labeled as duplicates as long as the questioner links to the original and briefly explains the differences and (in spoilers) how it might lead to different answers.
  4. Mistakes in questions that result in invalid questions may be edited as long as it follows rule 1.
  5. "Loopholes" included in answers will be allowed so long as they are based in logic. For example, strategies such as using cryptography to communicate publicly or paradoxes to trick knights/knaves/randoms are generally allowable, while strategies like threatening/beating up characters until they give you desired information are generally not allowed.
  6. While questioners should be as clear as possible in their questions (and should be prepared to receive votes for clarity as required), it is the responsibility of answerers to ask for clarifications on vague details prior to solving the puzzle. For example, in a knight/knave/random puzzle, if "a random" is not defined, the answerer should confirm whether "a random" randomly tells lies/truth, or if it randomly says yes/no.
    • a. Reasonable extrapolations made by an answerer do not fall under this rule. For example, if a character has a device, and the potential answer incorporates describing said device, there is no explicit need for a clarification. While it might be a good idea to get such details clarified, reasonable extrapolations being disallowed is the responsibility of the questioner, and should follow the rest of the rules in this guide.
    • b. In the case that the answerer made assumptions that were neither written nor implied because on vague statements in the question, the questioner must put in an edit clarifying the vague statements, following the rule 1.
    • c. While questioners should copy clarifications made in comments into their actual post (following Rule 1), answerers are responsible for following clarifications made in the comments.

That's all I can think of for now. Obviously people should be informed of these rules and asked to follow them before we get into downvotes and close votes and stuff like that.


I hope we can work together and solve this problem once and for all. What I've written above is my suggestions, and I would be happy to hear other people's opinions and suggestions as well. Thanks for taking the time to read this.

  • $\begingroup$ As for part of your solution - we can't edit most Help Center articles at all, let alone make a custom one. That stuff is unified across the network. Additionally, see Exit strategies for “chameleon questions” for a related main-meta discussion. $\endgroup$
    – bobble
    Aug 30 at 0:14
  • $\begingroup$ Some questions: by disallowing an edit, do you mean simply rolling the edit back and then escalating if the poster attempts to re-edit? (Which is what I assume, but I don't think it's made explicit). Also, for point 5. I assume that "loopholes based on logic" would be allowed unless explicitly disallowed? E.g. a liars puzzle saying "paradoxes are not allowed". $\endgroup$
    – bobble
    Aug 30 at 0:25
  • $\begingroup$ @bobble My mistake, I didn't realize Help Center was the same for everyone. I believe there is a page for Puzzling SE rules, maybe we could put something there? Also, both the things you said in the second comment is correct. $\endgroup$
    – Ankit
    Aug 30 at 14:50

You must log in to answer this question.

Browse other questions tagged .